N R Hardiker1. 1. Salford Health Informatics Research Environment, University of Salford, Manchester, UK. n.r.hardiker@salford.ac.uk
Abstract
OBJECTIVES: Several researchers have proposed the use of logical ontologies as 'reference terminologies'. However, there are a number of unresolved issues. This article describes the development of a logical ontology for nursing interventions and presents the results of evaluation. METHODS: Initially this study involved the development in GRAIL of two separate experimental ontologies: an ontology based on the textual content of informal definitions for nursing interventions drawn from the Nursing Interventions Classification; and an ontology based on labels for the same nursing interventions. Following initial bench-testing, the ontology based on labels was selected for extension (to accommodate also nursing intervention components of the Home Health Care Classification System and the Omaha System), for further testing and for external evaluation. RESULTS: A hierarchy of nursing interventions generated automatically from the experimental ontology based on informal definitions contained only 3 hierarchical relationships, compared to 214 for the initial ontology based on labels. For the final extended ontology based on labels, the generated hierarchy contained the three source terminology systems in entirety - there were a total of 2861 hierarchical relationships. While the results of comparative bench testing of the final ontology were favourable, the results of external evaluation were mixed and showed little agreement between reviewers. CONCLUSION: This study suggests that while a logical ontology based on labels might be a useful tool for mediating between nursing intervention terminology systems, a formative consensus type development methodology might improve the approach by helping to harmonise ideological differences that may exist across the nursing profession.
OBJECTIVES: Several researchers have proposed the use of logical ontologies as 'reference terminologies'. However, there are a number of unresolved issues. This article describes the development of a logical ontology for nursing interventions and presents the results of evaluation. METHODS: Initially this study involved the development in GRAIL of two separate experimental ontologies: an ontology based on the textual content of informal definitions for nursing interventions drawn from the Nursing Interventions Classification; and an ontology based on labels for the same nursing interventions. Following initial bench-testing, the ontology based on labels was selected for extension (to accommodate also nursing intervention components of the Home Health Care Classification System and the Omaha System), for further testing and for external evaluation. RESULTS: A hierarchy of nursing interventions generated automatically from the experimental ontology based on informal definitions contained only 3 hierarchical relationships, compared to 214 for the initial ontology based on labels. For the final extended ontology based on labels, the generated hierarchy contained the three source terminology systems in entirety - there were a total of 2861 hierarchical relationships. While the results of comparative bench testing of the final ontology were favourable, the results of external evaluation were mixed and showed little agreement between reviewers. CONCLUSION: This study suggests that while a logical ontology based on labels might be a useful tool for mediating between nursing intervention terminology systems, a formative consensus type development methodology might improve the approach by helping to harmonise ideological differences that may exist across the nursing profession.